This chapter is from the book
This chapter is from the book
This chapter is from the book
Geometry Shaders
The geometry shader is logically the last shader stage in the front end, sitting after the vertex and tessellation stages and before the rasterizer. The geometry shader runs once per primitive and has access to all of the input vertex data for all of the vertices that make up the primitive being processed. The geometry shader is also unique among the shader stages in that it is able to increase or reduce the amount of data flowing through the pipeline in a programmatic way. Tessellation shaders can also increase or decrease the amount of work in the pipeline, but only implicitly by setting the tessellation level for the patch. Geometry shaders, in contrast, include two functions—EmitVertex() and EndPrimitive()—that explicitly produce vertices that are sent to primitive assembly and rasterization.
Another unique feature of geometry shaders is that they can change the primitive mode mid-pipeline. For example, they can take triangles as input and produce a bunch of points or lines as output, or even create triangles from independent points. An example geometry shader is shown in Listing 3.9.
Listing 3.9: Our first geometry shader
#version 450 core
layout (triangles) in;
layout (points, max_vertices = 3) out;
void main(void)
{
int i;
for (i = 0; i < gl_in.length(); i++)
{
gl_Position = gl_in[i].gl_Position;
EmitVertex();
}
}
The shader shown in Listing 3.9 acts as another simple pass-through shader that converts triangles into points so that we can see their vertices. The first layout qualifier indicates that the geometry shader is expecting to see triangles as its input. The second layout qualifier tells OpenGL that the geometry shader will produce points and that the maximum number of points that each shader will produce will be three. In the main function, a loop runs through all of the members of the gl_in array, which is determined by calling its .length() function.
We actually know that the length of the array will be three because we are processing triangles and every triangle has three vertices. The outputs of the geometry shader are again similar to those of a vertex shader. In particular, we write to gl_Position to set the position of the resulting vertex. Next, we call EmitVertex(), which produces a vertex at the output of the geometry shader. Geometry shaders automatically call EndPrimitive() at the end of your shader, so calling this function explicitly is not necessary in this example. As a result of running this shader, three vertices will be produced and rendered as points.
By inserting this geometry shader into our simple one tessellated triangle example, we obtain the output shown in Figure 3.2. To create this image, we set the point size to 5.0 by calling glPointSize(). This makes the points large and highly visible.
)
Figure 3.2: Tessellated triangle after adding a geometry shader